Environmental Correlates of Physical Activity And Walking in Adults and Children: A Review of Reviews

Transcription

1 Environmental Correlates of Physical Activity And Walking in Adults and Children: A Review of Reviews A.E. Bauman and F.C. Bull Review undertaken for National Institute of Health and Clinical Excellence February 2007 Contact Dr Fiona Bull Physical Activity and Health BHF National Centre for Physical Activity and Health School of Sports and Exercise Science Loughborough University Tel +44 (0) f.c.l.bull@lboro.ac.uk 1

2 EXECUTIVE SUMMARY The purpose of this report is to summarise evidence on the environmental correlates of physical activity and walking. A review of reviews was undertaken to identify evidence on the strength of association between environmental factors and physical activity for both children and adults. The exposure variables were a range of measures reflecting elements of the physical environment, and the outcomes of interest were total physical activity and specifically walking behaviour. Electronic searches were conducted using Medline Cinhal, ERIC, Psychlit and Sportsdiscus databases as well as hand searches of references list and selected key journals. The focus was to identify published papers reporting on the factors associated with physical activity in children and adult populations; these are referred to as correlates or determinants studies. The initial search identified over 200 papers, of which only 13 papers met the inclusion criteria and were included. These review papers varied in quality and approach taken to reviewing and presenting their results. Despite these variations some common themes were identified across the studies in terms of both results on environmental correlates and the methodological weaknesses of this field of research. A summary table presents the key environmental factors that show consistent and statistical association with physical activity behaviour in adults. For physical activity, 11 review papers provided evidence and results across them revealed reasonably consistent associations between access to physical activity facilities, convenient and proximate access to destinations, high residential density, land use and urban walkability scores. There were also reasonably consistent associations between perceived safety, exercise equipment, pavement ( sidewalks ) and physical activity participation. Less clear associations were noted for aesthetic features of the environment, parks, and perceived crime. Four review papers provided evidence specific to walking and the results showed the correlates of walking were similar to those found for physical activity in adults. One review paper assessed the evidence on the environmental correlates of physical activity in children, it included 33 primary studies. The most consistent associations were for the provision of pavements ( sidewalks ), destinations to walk to, few intersections to cross and low road traffic hazards. Aspects of the recreation infrastructure were also found to be strongly associated with increased levels of activity, these included proximity to, and availability of to parks, playgrounds, and recreation areas. Less consistent results were found for perceived safety however some studies show crime rates may be significantly (negatively) associated with physical activity in children. 2

3 The literature for both adults and children populations was mostly from cross sectional study designs, most studies were conducted in USA or Australia using a variety of outcome and exposure measures. There were few clear differences in the results across the studies included to identify patterns by age, sex or ethnicity, in adults or children Limitations in this literature include the reliance on cross sectional study design, thus this temporal relationship is not known for environmental measures and it is not possible to infer a causal or true determinants relationship. The lack of standardisation of measurement of environments is another limitation. Both the independent and dependent variables in this area of research pose measurement challenges. A wide variety of methods are used to collect, derive, analyse and present data. This makes the task of assessing and comparing the results across studies more difficult. Few reviews identified separately the findings by objective or subjective measurement methods yet this will remain an important issue. Future research should make an explicit attempt to do so. Limited research is available on correlates of physical activity in minority and specific sub-populations. Research on correlates has been carried out almost exclusively in North American and Australia, with limited published literature on environmental correlates from the UK. Thus, the evidence base for understanding the role of the environment and physical activity in England is very limited. Until primary research studies are undertaken in the UK, the application of the findings to date will rely on the generalisability of the available literature. Limitations due to the methods of this review may include limiting the review to a review of reviews, including studies published only in English between and applying search terms to only title and key words. 3

4 1. INTRODUCTION 1.1 Scope of the review The purpose of this report is to summarise evidence on the environmental correlates of physical activity. In this relatively new field of study of physical activity, the evidence most often comes from cross sectional studies rather than cohort or intervention research designs. The results will present a summary of identified environmental factors that show consistent and statistical association with physical activity behaviour in adults and children from a review of the published literature. Given the publication of well over a hundred papers addressing association between the environment and physical activity and the time limitations of this project, the evidence presented is derived from review level evidence and presents a narrative summary of the strength, consistency and direction of these associations. As there are different ways that exposure and outcome factors have been measured, a formal synthesis and quantitative comparison is not possible, but the objective, to provide a narrative review, will inform policy makers on the strength of the evidence to date. The results section of this report is divided into several sections covering firstly the literature identifying the environmental correlates of physical activity and secondly the environmental correlates of walking. For both outcomes the available evidence for both children and adults is covered and, where possible, the results are presented to show results from objective measures and subjective perceptions separately. An additional final section provides commentary on the measurement methods used in this field of research. The overall purpose of this review is to provide a synthesis of the available evidence on the potential environmental determinants of physical activity. An understanding of these factors can inform the development of more effective interventions and direct policy action that will assist in the development of promising areas for intervention to change the environment which may positively influence physical activity at the population level. 1.2 Summary of health benefits of physical activity Physical inactivity has been identified as one of the leading causes of many chronic conditions, and it contributes substantially to the burden of disease, disability and premature death, with heavy 4

5 resulting economic costs (Department of Health, 2004). Recent efforts to promote and increase physical activity at the population level have focussed heavily on individual and behavioural interventions and yet rates of physical activity remain relatively static and at the same time sedentary activities have increased. This has contributed to increasing non-communicable diseases, and partially contributed to the rise in obesity rates. Understanding the correlates and determinants of physical activity is essential in the development of a comprehensive population-based approach to the promotion of physical activity. 1.3 Conceptual Model for Understanding Correlates and Determinants of PA The socio-ecologic model posits that comprehensive approaches to change physical activity levels need to consider interventions at multiple levels, individual, social and environmental. The latter component, environmental influences on physical activity, is the subject of this report. As a new field of research, a large number of different characteristics of the environment have been identified and explored. These include single and aggregated constructs, at varying levels of scale (local neighbourhood such as a few streets through to characteristics assessed at the suburb or city level) and can include: density of shopping facilities; distance to destinations; mixed land use; level of connectivity; provision of walking and cycling facilities; street lighting; access to open spaces and population density. Consideration has also been given to objectively assessed characteristics as well as subjective (perceptions) assessments of the environment (such as perceived safety, perceived levels of traffic, aesthetic qualities, convenience of access to facilities). Two examples of socio ecologic models adapted to apply to physical activity are shown in Figure 1. There is however no consensus on any one specific ecologic model although more recent work has called for greater specificity in the development of models for specific types of physical activity outcome variables (Giles-Corti et al., 2006). 5

6 Figure 1 Examples of Socio-Ecologic Models applied to Physical Activity A) Example 1: Spence and Lee, B) Example 2: Pikora et al., Spence J. C. and Lee, R.E. (2003) Towards a comprehensive model of physical activity Psychology of Sport and Exercise. 4: Pikora,T., Giles-Corti, B., Bull, FC. Jamrozik K., Donovan, R. Developing a Framework for Assessment of the Environmental Determinants of Walking and Cycling. Social Science and Medicine. 2003: 56:

7 1.5 Individual and Social Correlates of Physical Activity Much of the available evidence on correlates of physical activity has been collected at the personal or individual level. A recent review was undertaken for the World Cancer Research Fund by the authors of this report and the main findings are summarised in Table 1 (Bauman and Bull, In press). Among adults, there are consistent demographic factors associated with physical activity in populations. These include age and gender, with most leisure-time physical activity measures showing lower prevalence of physical activity among women and older adults, compared to men and young adults. In addition, some socio-economic individual level attributes, especially educational attainment and income are positively related to leisure-time physical activity. 3 One of the most consistent intra-personal factor consistently associated with physical activity is self efficacy, part of social cognitive theory, and is related to the person s situation-specific confidence that they can be physically active. Self-efficacy and related constructs are measured frequently, and remain positively associated with physical activity across diverse population groups and populations. Other intra-personal variables, including attitudes and beliefs are inconsistently associated with activity. Perceived barriers to physical activity especially perceived lack of time [adults] and feeling too old or having chronic illness [older adults] were consistently and inversely related to activity. Exercise habit or a history of previous participation earlier in adult life were identified as correlates of current adult participation. In adults, social support is associated with physical activity, particularly among older adults, women, and some minority populations. Cultural factors have some influence, but it is not consistent and is an area in need of more research Among children and adolescents, consistent associations were noted for sex [males more active], and for age, with declines in physical activity through adolescence. Some intrapersonal perceptions, such as body image and self perceptions [inverse] and self-esteem [direct] are associated in some studies. 3 Note that other relationships may exist for other types of physical activity for example active commuting may show an inverse association with socio economic status, as those of least income may not be able to afford cars and other sedentary transport modes. 7

8 Table 1 Summary of the coverage of Inter- and Intra- individual, barriers and social correlates of physical activity in adult populations* Trost et al Seefeldt et al., 2002 Recent Review Papers White Sherwood et al., & Jeffery Speck et al., 2003 Demographic and Biological Factors Age Gender Indicators of socio-economic status: educational attainment, occupational status and measures of SES) Ethnicity/race Dependent children Marital status Year of study in higher education Overweight or obesity or body weight General health (or health status) Level of fitness Functional status Mental wellbeing Psychological, Cognitive, and Emotional Factors Attitudes Barriers Enjoyment Expected benefits Value of outcomes Outcome expectancy Intentions Normative beliefs Self efficacy Self motivation Stage of change Self efficacy Barriers Lack of time Too weak (elderly populations) Fear of falling (elderly populations) Bad weather No facilities Lack of exercise partner Lack of energy Self consciousness Fear of joint pain / MI / chest pain Injury Attitudes Knowledge Adherence related correlates Mode (type) of physical activity Duration of activity sessions Setting (home / workplace) Choice / variety Behavioural attributes and skills Diet habits Past exercise behaviour Past sports participation Smoking status Decisional balance Stress levels * Adapted from a recent review by Bauman and Bull (In Press) Keating et al.,

9 Intention to be active and leisure time preferences are important in younger children, but not apparent in adolescents. Previous physical activity and membership of community sports associations were associated with being active in adolescence. Among younger children, social influences are quite important. Consistently, parental and other adult role models are important, but this becomes less important among adolescents. Peer influences and teacher modelling of active behaviours is unrelated to youth physical activity. 1.4 Outcome Variables: Physical activity and walking This review will consider environmental correlates of total physical activity as measured in the studies and usually this involves self report instruments that address a range of domains of activity, such as, leisure, transport and possibly at work. Correlates of specific behaviours, namely walking, will also be identified, where possible from the literature. The term exercise may be used in this report as this is still a common term in the relevant literature base. Whilst the terms are often used interchangeably in studies, it is important to note that the concepts of physical activity and exercise do differ, with the former being a broader term, the latter a term used to describe activity undertaken in usually more structured ways and with specific purpose of improving aspects of fitness (e.g., cardiovascular, muscular strength, flexibility). 1.5 Correlates versus determinants of physical activity A synthesis of evidence on the determinants and correlates of physical activity is presented. Strictly speaking, the focus here is on the correlates, as in measures of associations that are consistently reported within the data, rather than results on antecedent causal variables [or determinants]. This field of research has a history of using both terms and often (incorrectly) interchangeably; it is only more recently that the strict use of the terms has been evident in published papers. In this review we incorporate all evidence that is, in the epidemiological sense conceptualised as correlates of physical activity to identify those environmental factors consistently and statistically associated with physical activity behaviour. 9

10 2 METHODS This report summarises the evidence on studies that have examined the relationship between physical activity or walking and aspects of the built or physical environment. This report uses a review of reviews approach, with the following steps undertaken: i. A search strategy prepared and undertaken ii. Assessment of hits for relevance and inclusion iii. Synthesis of each included review paper to provide a narrative description, considering, where possible, all the following: a. Research purpose b. Search strategy and timeframe c. Number of studies included and their study designs d. Main findings from each review e. Main conclusions from each review f. Implications for policy iv. An additional focus was included to identify the measurement methods used in the papers reviewed, and thus provide a general discussion of measures of levels of physical activity or walking, and measures of environments, and where possible, to identify any differences in findings between results and conclusions based on studies using self report and objective measurements v. Assessment of robustness and / or weaknesses of the evidence base in this area. vi. Consideration of these findings in terms of the context and relevance for the UK vii. Identification of future research needs and directions to enhance the evidence base for action Search strategy An electronic search was undertaken to identify papers reporting on factors associated with physical activity in children and adult populations; these are referred to as correlates or determinants studies. Given this diverse and large body of evidence, and in order to undertake a review in the timeframe provided, review-level evidence was sought, with a preference for systematic methodologies. Thus, the search strategy focused on identifying (quantitative and qualitative) review papers with a focus on the correlates of physical activity that emphasized the built environment or physical environment. Studies of individual level and social environmental 10

11 correlates were excluded. In addition, the review was limited to papers published in English language and published between Searches were conducted in Medline and four other data bases (Cinhal, ERIC, Psychlit, SportDiscus) and separately run in ISI Web of Science. In addition, the authors undertook hand searching of reference lists, key journals and used their network of researchers working in the area of physical activity and the physical environment to identify additional studies. 5 Search terms used were physical activity or walking, correlates or determinants, environment* or built environment or physical environment as key words or title words (Table 2). In addition, ISI Web of science was examined, for physical activity (topic search or title search along with environment as a topic word or title work, OR title words of ("public health" OR ecolog* OR transport* OR "urban planning" OR "urban design" or transport*). Table 2 Category Search terms for review of physical activity and environments Search terms used Physical activity Review paper selection Environment Association Physical activity or exercise or walking in title or key word Defined as a review paper or with key word or title word of [any of] review /systematic /research synthesis /summary environ$ or built environment or physical environ$ in title, or keyword correlates or determinants title or key word Reviews were excluded if their focus was not on physical activity but rather studies of correlates of solely strength training or clinical exercise programmes (such as exercise for rehabilitation), if the paper reported results of a single study; or if there was no clear evidence of a review or summary process. In addition, papers were excluded if the focus was on a single disease or specific clinical setting as this report aimed to assess the correlates of physical activity that might be of potential population or public health usefulness (Table 3). Four reviews of interventions studies on the environment and physical activity were identified but excluded (McCormack et al., 2004; Owen et al., 2004; Saelens et al., 2003; Wendel-Vos, G.C. 2004). 4 The reason for the recent time frame is that all the most useful reviews have been written in the past five years because more than 80% of the primary source studies have been published in the period since 2002 (Wendel Vos et al., 2005). 5 Note that the three separately conducted search processes identified some of the same review papers this overlap is useful as it utilises a triangulation approach to optimise bibliometry and collect a more comprehensive set of reviews. 11

12 Table 3 Inclusion criteria for Review papers on physical activity and environments Criteria Study focus and design Study included if: i. Was classified as a review paper ii. Used a clear measure of physical activity (PA) or walking specifically; studies were excluded if they only examined other aspects of physical activity, such as fitness, exercise training or strength training, and if they related the PA measures to studies of the individual, social or behavioural correlates, although these are referred to in the discussion section iii. There was evidence of a review or summary process; single studies of the associations between physical activity and environments in particular populations or samples were not included; iv. Was published in English, in Table 4 Summary of search results by method obtained or data base searched [note results in each column not independent] Stage 1 All papers identified using search terms in Medline and in other data bases* or ISI Web of Science Medline or any of the five * data bases # identified 2 All Review papers 58 classified as reviews 3 Reviews for further consideration ISI web of science Current Contents papers extracted # identified Personal communications or reviews carried out by authors research teams # identified primary papers for PA and environments; 72 for walking and environments identified 12 classified as reviews reviews identified Note: These sources produced several reviews in common; the final 13 chosen for direct inclusion in this report were derived from the three sources together [Medline, ISI web of science and personal sources]. * the five databases : Medline Cinhal, ERIC, Pyschlit, Sportsdiscus # note that review papers not available from other databases directly search term included review, systematic, research synthesis, summary A total of 192 hits were identified on Medline and related databases and 148 hits were identified in Current contents ISI Web of Science; additional papers were manually considered from personal communications and networks, and from the reviews conducted or in progress by the authors research groups to further extend the search. After reviewing for relevance and against the inclusion criteria, a total of 13 reviews were included. 12

13 3 RESULTS The search strategy identified 13 review papers that explored the relationship between physical environments and physical activity or walking. Eleven of these focused on identifying environmental correlates of physical activity. One review focused exclusively on walking and three reviews provided information on both physical activity and walking. Only one review specifically examined the correlates of activity in children and adolescents (Davidson et al., 2006). Included studies all met the search term criteria, and were classified by the electronic databases and by the authors of the papers as a review paper. All extracted some studies and commented on the overall pattern of the relationships seen between environments and physical activity; in other words, they were quantitative studies. However, the quality of the review methods and process, for example in terms of conducting systematic searches and providing the details and numbers of studies obtained and reviewed, were not consistent across the reviews. Conducting a formal quality audit of this set of reviews was beyond the scope of this review of reviews, but the variation in quality across the review papers made it more difficult to find and report on all aspects of interest. Nonetheless, efforts were made to extract details from each review that were comparable in terms of research purpose, search strategy and timeframes, main findings and implications for policy. Table 5 presents the review papers included in this review on environmental correlates of physical activity. Below is a descriptive synopsis of each of the included review papers with an emphasis on the providing information on the following sections (where relevant details could be extracted from the review): the main purpose or research question; search strategy and databases used; the studies included, and in turn, their study design and methods; estimates of association; and country and/or context (population). Review papers are presented in alphabetical order. 13

14 Table 5 Reference of Included Review Studies 1 Badland, H., Schofield, G. (2005). Transport, urban design, and physical activity: an evidencebased update. Transportation Research Part D: Transport and Environment, 10(3), Cunningham, G. and Michael, Y. (2004). Concepts guiding the study of the built environment on physical activity for older adults: a review of the literature. American Journal of Public Health, 18(6), Davidson K.K. and Lawson C. (2006) Do attributes of the physical environment influence children s level of physical activity? International Journal of Behavioural Nutrition and Physical Activity. 3 (19) Duncan, M. Spence and Mummery K. (2005) Perceived environment and physical activity: a meta-analysis of selected environmental characteristics. International Journal of Behavioural Nutrition and Physical Activity 5; Humpel, N., Owen, N. and Leslie, E. (2002). Environmental factors associated with adults participation in physical activity. American Journal of Preventive Medicine, 22(3), Lee, C. and Moudon, A. (2004). Physical activity and environment research in the health field: implications for urban and transport planning practice and research. Journal of Planning Literature, 19(2), McCormack, Giles-Corti, Lange A. Smith T. Martin K, and Opikora T. (2004) An update of recent evidence of the relationship between objective and self report measures of the physical environment and physical activity behaviours. Journal of Science and medicine in Sport, 7 (1) Owen N, Humpel N, Leslie E, Bauman A, Sallis JF. (2004) Understanding environmental influences on walking: Review and research agenda. American Journal of Preventive Medicine. 27(1): Saelens, B., Sallis, J. and Frank, L. (2003). Environmental correlates of walking and cycling: findings from the transportation, urban design and planning literatures. Annals of Behavioural Medicine, 25(2), Sallis, J., Frank, L., Saelens, B. and Kraft, M. (2004). Active transportation and physical activity: opportunities for collaboration on transportation and public health research. Transportation Research, 38, Trost S.G., Owen N., Bauman A.E., Sallis J. F., Brown, W. (2002) Correlates of Adult s participation in physical activity: Review and Update. Med Sci Sports and Exercise 34 (12) Vojnovic, I. (2006). "Building communities to promote physical activity: A multi-scale geographical analysis." Geografiska Annaler Series B-Human Geography 88B(1): Wendel-Vos, G.C. Schuit, A. Jantine ; De Niet R. Boshuizen H Saris W. Kromhout D. (2004) Factors of the physical environment associated with walking and bicycling. Medicine & science in Sports & Exercise. 36(4):

15 3.1 Environmental correlates of physical activity among adults 1. Badland and Schofield (2005) This review paper provided a narrative description of earlier review papers and then described the origins of the study of urban form that were potentially related to physical activity. The primary studies were included if they examined differences in neighbourhoods and physical activity [n=13] or factors associated with non-motorised transport [n=11 studies]. Neighbourhood difference factors revealed significant correlates with activity. These included residential age of houses, 6 walkability, 7 mixed land use, street connectivity and the number of intersections, access to facilities within walking distance, parks, lower traffic and public transport. Factors associated with non-motorised transport usage included shorter distance to destinations or transport, gender, social status, aesthetic factors and access to public open spaces. Badland and Schofield (2005) indicated that the limitations of the field include a lack of causal evidence, as well as the limitations imposed by reviews focusing on academic publications only; the grey literature may have additional information of use to policy makers. This is especially true around travel choice behavioural studies, few of which are in the scientific published literature. 2. Cunningham and Michael (2004) This review focused on studies [particularly in older adults] that explored the relationship between physical environments and physical activity. The Cunningham review identified studies published between , and 27 papers in English were located. Of these, 6 studies were specific to older adults. A strong emphasis was to identify theoretical frameworks for these studies [found in 17 papers]. Travel demand theory, environmental press theory, and socio-ecologic models were identified. All except two studies used representative population samples. The correlates were not grouped or categorised, but described by author and year of publication. In this review, one third of the 65 estimates of association reported in these studies were non significant. The correlates that generally showed significant associations with physical activity were 6 the era when houses were built is a proxy measure for medium density, as increased urban density in planning housing developments characterised suburbs before the 1960 s 7 Walkability is the sets of physical environmental attributes associated with increased probability of walking there are no agreed definitions of walkability; this is because the environmental factors associated with walking are only now being distilled in review papers. One example is Saelens et al., (2003) who propose and report on a validated measure of walkability that includes measures of land use mix, density, and connectivity. 15

16 neighbourhood safety, footpath safety, access to convenient facilities, enjoyable scenery, shorter distances to facilities or parks, low perceived crime rates and low traffic density. Similar correlates applied to older adults, with a specific emphasis on convenience, access and safety. The authors identified the limitations of the literature in terms of study design and measurement. Cunningham and Michael asserted that limited inferences could be made from studies using crosssectional designs, and further work is needed to characterise and assess the environment factors that may be relevant or older adults. Ageing-specific frameworks for this research were recommended. 3. Duncan, Spence and Mummery (2005) This paper is the only attempt at a quantitative review of the association between (perceived) environmental attributes and physical activity. The authors describe the need for a quantitative review because the associations between the perceived environment and physical activity are ambiguous (Duncan et al., 2005; p2). They proposed a formal meta analysis and identified papers through Medline, Proquest, Infotrac between 1989 and 2005, along with a manual search of five key identified journals. They found 50 candidate studies, with 138 estimates of environment physical activity associations. Studies were screened for inclusion, based on populations chosen, independent samples, and if multiple environmental attributes were considered. Sixteen primary studies were included and effect sizes were calculated and compared, albeit from slightly different exposure and outcome measures. The authors examined the presence of covariate data, such as gender, country of research (93% were conducted in the USA), rurality, and coded for type of physical activity or walking behaviour that was measured. The meta analytic effect size (ES) results were summarised as pooled odds ratios with confidence intervals. The results showed that unadjusted, no pooled odds ratios from the meta-analysis were significant, but adjusting for age, education and gender provided some evidence of significant effect sizes. Those that reported physical activity facilities in the neighbourhood were 20% more likely to be active, the presence of pavements increased the likelihood of being active by 29%, and shops and services in walkable distance increased the likelihood of activity by 30%. Absence of heavy traffic also increased the likelihood of physical activity by 22%. These results were significant however, the adjusted pooled odds ratios for perceived rates of crime, street lighting, and unattended dogs were non significant. The authors opined that these effect sizes might be different for different subgroups, but there were not enough primary studies to assess the veracity of this hypothesis, but it is 16

17 of future research importance, in refining our understanding of these correlate studies. This issue is also raised in the review of Wendel Vos et al (2005). The authors (Duncan et al.,) used estimation techniques to approximate the percent of variance in physical activity explained by each environmental attribute. The results showed R 2 ranging from values of 4% for traffic to 7% for proximity to shops and services. This is similar to the estimates to total environmental contribution to physical activity among adults (Sallis et al., 2000) and among children (Fein et al., 2004). This meta-analytic study provides evidence that physical environments are related to the presence of physical activity facilities, proximate destinations, pavements and perceiving traffic not to be a problem. These findings were based only on self-reported perceptions of the environment, but are consistent with other reviews described, especially one of the earliest and comprehensive reviews by Trost et al., (2002), and the careful review also published in 2005 by Wendel Vos et al., (2005). 4. Humpel, Owen, Leslie et al., (2002) This comprehensive systematic review identified 19 studies, of which 16 described the relationship between physical environments and physical activity. This review only focused on physical activity behaviours as the outcome variables of interest. Four of these studies reported an objective measure of the environment, eleven used self-report measures and one used both objective and self-report. Fifteen of the 16 studies were cross sectional in design. A secondary assessment of the studies in this review indicates that 14 observed associations in large representative population samples. The factors positively associated with physical activity include accessibility of facilities [cycle path; facilities, nearby park or beach or shops in walkable distance], and opportunities for activity [home equipment; satisfaction with facilities; local clubs that provide opportunities to be active; coastal residence]. Factors that reduce the likelihood of being active include hills, busy traffic, low density residential regions, no facilities nearby, and a lack of equipment. There were some aesthetic factors associated with increased physical activity such as the perception of a friendly, pleasant and attractive neighbourhood. In addition, other factors such as safe pavements, low perceived crime rates were also associated with physical activity; however, weather, unattended dogs, presence of streetlights and daytime perceived safety were not related to activity levels. 17

18 Implications of this review were to reinforce the consistency of observations regarding the physical environment. Objective measures of the environment, such as using geo-coding through geographic information systems [GIS] seemed to be differently related to some perceived [self report] environmental questions. The authors argue against the summation of physical activity items into total environmental measures or scores, as the different elements might relate in different or even opposite ways with respect to physical activity. 5. Lee and Moudon (2004) This paper, written by urban planning academics, and published in the planning literature, provides a descriptive summary of urban form features identified as associated with health enhancing physical activity. 8 A diverse range of environmental exposures were assessed. These included locations where people preferred to be active were assessed, and local streets, parks and sports fields, public open space (including beaches), specific trails and paths, and shopping malls were the most frequently reported. Research identifying environmental barriers to physical activity was categorised as lack of access to facilities, travel distance and safety concerns. The authors reviewed three electronic databases (Medline, Psychinfo, ISI web of science), as well as the grey literature from federal and state agencies. The literature review was conducted during , but the number of initial candidate studies was not stated. Of these candidate studies, the authors reported findings from 20 studies, some of which used qualitative research methods or did not report quantitative data. These authors divided these correlational studies into those that assessed environments objectively, and separately described those that self-report measures of environments. They developed a theoretical framework, described as the behavioural model of the environment, considered route specific attributes and origin/destination specific attributes of the physical environment. The results showed that objective measures of environments generally assessed objective characteristics of origins / destinations [such as distance to the shops], and route-based measures [such as the presence of a bike lane or footpath] were often assessed by self-report measures. Some geographical area level variables, [e.g. coastal residence, age of housing, hilliness] were examined as proxy indicators of general urban form characteristics. 8 The paper identified the prevalence rates of walking and cycling for different purposes, including active commuting and activities of daily living. Walking for transport was often reported by Australian and Canadian adults with a quarter of adult Australians reporting walking and <10% cycling for transport in the previous week. Among Canadians, 26% biked for transportation and 85% walked for transport at least sometimes. 18

19 Lee and Moudon concluded that destination availability and quality were related to usage. Other ecological factors, such as coastal proximity, urbanisation, density and convenience were also related to physical activity levels and usage. In studies with self-report measures of environmental attributes, local neighbourhood features and quality, the presence of destinations, access to facilities and the existence feasible routes, as well as other aspects of walkability were related to undertaking sufficient physical activity for health. Perceived safety was also positively associated with physical activity levels. This review concluded that there was a latent or unmet demand for transport-related walking and biking, and that these activities were in turn related to macro-scale environmental variables such as urban density, land use, street design, and the availability of destinations. The authors suggest greater investments in public transportation, facilitating direct active commuting through better pavements and bike lanes, and other activity and environment-friendly improvements. 6. McCormack, Giles-Corti, Lange et al., (2004) This review paper focused on the built environment and identified papers published since Only quantitative studies were included, with self-report or objective measures of the environment. The theoretical framework used was one that was previously developed by the authors, and was comprised of functional attributes of the physical environment, safety aspects, aesthetic aspects, and destinations. The total number of studies considered is around 30, but is not explicitly stated. From these studies, 20 significant estimates of association with physical activity measures were reported, and a further 15 significant estimates of correlates with walking. McCormack et al., identify that functionality, in terms of environmental sprawl, lower population density, and low walkability were associated with lower rates of physical activity. An overall sprawl index was related to walkability and to activity levels. Across the included studies, safety was usually assessed using self-report measures and showed inconsistent relationships to activity, but pavement safety seemed to be regularly associated with higher activity levels. Self reported measures of aesthetic factors, such as cleanliness, scenery, varied building designs, low traffic and greenery, were found to be positively associated with activity. None of the included studies assessed the aesthetic factors using objective measures. Topographic factors were inconsistently related; sometimes, hilly terrain and busy roads and freeways were a barrier to activity, but these were not 19

20 consistent results. Destinations, in the form of access to facilities, proximity to walking and activity opportunities and shops were positively associated with activity. The review concluded with the limitations imposed by self report and by cross-sectional designs in producing evidence relevant for public health policy. The needs for better [causal] designs, and objective measures of both physical activity and environments is recommended. 7. Saelens, Sallis and Frank (2003) This review summarised studies that examined neighbourhood characteristics associated with walking or cycling for transport. The theoretical underpinnings were socio-ecological models, that include intra- and interpersonal factors, as well as social and physical environments in contributing to physical activity participation. This review summarised data from four studies that identified higher rates of walking or cycling for transport in neighbourhoods that were deemed highly walkable, compared to low walkable regions. The environmental attributes that comprised walkability included population density, mixed land use [both residential and non-residential purposes in the same region], and the presence of facilitatory infrastructure [bike lanes, pavements]. Selection of studies was not clear, but the authors examined the health, transport and urban planning literatures. The differences between high and low walkable regions were most marked for transportrelated physical activity, more than for leisure time exercise behaviours. The policy relevance of this review was to highlight the potential contribution that transport related physical activity might make to reaching recommended levels of total physical activity for health (USDHHS, 1996). In addition, transport related activity can contribute to overall obesity prevention, through the increased and habitual energy expenditures attributable to these regular activities. 8. Sallis, Frank, Saelens and Kraft (2004) This paper described the relationships between community design and active transport. Much of this review described the principles of active commuting, and the concept of travel distance. Studies conducted by the authors had described the trips people made for different purposes, and that some trips to school, work or for shopping were less than a mile, but still undertaken using motorised transport. The authors then describe a systematic review of land use and active transport research. Theoretical frameworks included the socio-ecologic model, and also travel demand theories. 20

21 This research is based on the concept of walkability, a construct comprised of environmental characteristics, including population density, land use mix, and urban infrastructure [including street connectivity, and street amenities, such as well maintained pavements]. Several research projects have noted that walking trips were more prevalent in high walkable neighbourhoods, compared to low walkable neighbourhoods. The new studies reviewed here were six studies of active commuting and health outcomes, and were conducted in China, Japan, France, Ireland, Denmark and the Netherlands. Increased transport-related physical activity was associated with improved lipid profiles, lower body mass index (BMI), reduced hypertension rates, and lowered risks of allcause mortality. The effects on BMI seemed stronger for males than for females. The authors describe additional outcomes of increased active transportation at the population level. These include improved air quality and reduced traffic congestion. It was suggested that solutions to low rates of transport-related activity pose an intersectoral set of problems, and require transdisciplinary solutions, beyond simply work from within the health sector alone. 9. Trost, Owen, Bauman, Sallis and Brown (2002) This was a systematic review that provided an update of previous summaries of the correlates of physical activity in adults (Sallis and Owen,1999). In the earlier 1999 review, Sallis and Owen concluded that adult participation in physical activity was influenced by a diverse range of personal, social, environmental factors. This 2002 review used the same search terms as in 1999, and identified 38 new studies that were published since Seven of these used a longitudinal design, none used objective measures of physical activity and all relied on self-report. Most measures provided a global assessment of physical activity with the majority focusing on measures of leisure time activity only. Only 3 studies specifically focused on adherence to a structured exercise programme and these were retained in the pool of studies. The review summarised demographic factors associated with physical activity, including age, gender, educational attainment, and measures of socio-economic status. The detailed findings of these psychological, cognitive, and emotional factors assessed in this review are beyond the scope of this report. 9 The new part of this review was to add Physical Environmental Factors as a new 9 In brief, behavioural attributes and skills, such as previous habits, past exercise behaviour, smoking status and decisional balance for activity, were associated with participation in physical activity. Social support was consistently and positively associated with levels of physical activity. Overall the variance in physical activity explained by measured correlates ranged from 3-50% with an average 21.2 ± 15%. 21

22 set of correlates, identified in the published literature for the first time in studies since Ten studies generated the correlates between aspects of the physical environment and physical activity. Factors that emerged as potentially showing an association were: [i] exercise equipment in the home and [ii] access to and satisfaction with facilities; [iii] neighbourhood safety, [iv] a hilly terrain, [v] frequent observation of others being active (modelling) and [vi] enjoyable scenery. Urban/rural location was explored in 6 studies and all found physical activity to be significantly lower among adults living in rural areas compared with urban. Factors not found to be associated with physical activity in these studies included street lighting, programme costs, crime rates, sidewalks (pavements), and unattended dogs. This was one of the earliest review papers (published in 2002) and made a significant contribution to setting the agenda for the investigation of environmental correlates of physical activity, as well as demographic factors associated with being active. Although the notion of environmental influences on physical activity had been suggested well before this review (Nurmela, 1995) it was around this time that the evidence base started to accumulate. A major limitation of research published up until this point was a reliance on measures of leisure time physical activity, and not examining correlates of other measured domains especially transport and lifestyle-related physical activities. The other major limitation identified this review paper was the reliance on cross-sectional correlates, rather than longitudinal studies or interventions to assess the causal relationship between environmental factors and becoming physically active. 10. Vojnovic (2006) This review emanates from the planning literature, and it contributes a few innovative ideas. In particular, it proposes that the urban form and environmental correlates may be different at a microscale environment [local neighbourhood factors that encourage walking to the shops] compared to macro-scale environments [city design factors that encourage the development and use of public transportation systems, and facilitate active commuting to and from this public transport]. The review is not systematic, in that the sampling frame and selection of studies is not described. It is included here as it emanates from a planning perspective, and the theoretical and conceptual ways in which studies are described is quite different to other reviews. The environmental characteristics related to walking or cycling choices are characterised as personal factors (including perceptions of distance, safety, convenience, monetary cost, and time taken, as well as traditional intra-individual 22